I love your multilevel thinking! Very interesting. Indeed, things like decision latitude and other markers of empowerment/control could be context dependent and would be interesting to study.
Another option for modeling of SES is confounding, although, as we talked about in class, this is complicated because the direction of the causal arrow from SES to race/ethnicity does not make sense - so it is really more of a "third variable" whose treatment depends on what your research question is (e.g. do you want to know what the race/ethnicity association is independent of certain SES variables, acknowledging they are usually imperfect measures, or do you want to know the association of an outcome with race/ethnicity without controlling for the mediating effect of SES).